Re: 回复: [PATCH v1] arm64: dts: amlogic: meson-s4-s905y4-khadas-vim1s: enable SDIO interface

[neil.armstrong@xxxxxxxxxx]

On 1/20/26 02:34, Nick Xie wrote:
> [...]
> > > +&sdio {
> > > +       status = "okay";
> > > +       pinctrl-0 = <&sdio_pins>;
> > > +       pinctrl-1 = <&sdio_clk_gate_pins>;
> > > +       pinctrl-names = "default", "clk-gate";
> > > +       #address-cells = <1>;
> > > +       #size-cells = <0>;
> > > +       bus-width = <4>;
> > > +       cap-sd-highspeed;
> > > +       sd-uhs-sdr50;
> > > +       sd-uhs-sdr104;
> > > +       max-frequency = <50000000>;
> > > +       non-removable;
> > > +       disable-wp;
> > > +
> > > +       amlogic,dram-access-quirk;
> > Have you tried without this property? To my knowledge only early G12A
> > and G12B SoCs required this.
> > The mode enabled by this property limits the SDIO throughput. So it
> > should only be used if the silicon requires this.
>
> Without this property it will get an warning:
>
> meson-gx-mmc fe088000.mmc: unaligned sg len 96 blksize 512, disabling descriptor DMA for transfer
>
> > > +       no-sd;
> > > +       no-mmc;
> > > +       mmc-pwrseq = <&sdio_pwrseq>
> > > +       vmmc-supply = <&vddao_3v3>;
> > > +       vqmmc-supply = <&vddio_ao1v8>;
> > I'm missing:
> >    clocks = <&sdio_32k>;
> >    clock-names = "lpo";
>
> Maybe it is not needed? This clock property looks for the bluetooth not for the Wi-Fi ?
>
> > > +       brcmf: wifi@1 {
> > > +               reg = <1>;
> > > +               compatible = "brcm,bcm4329-fmac";
> > Is AP6256 using a bcm43456 chip? If so then this should be (according
> > to Documentation/devicetree/bindings/net/wireless/brcm,bcm4329-fmac.yaml):
> >     compatible = "brcm,bcm43456-fmac", "brcm,bcm4329-fmac";
>
> OK, I will fix it in next version.
>
> With this SDIO patch, the SDIO card detected well, but when try to connect the wifi, it still failed.
> So maybe extra patches are needed to make Wi-Fi working.
>
> [    5.909858] meson-gx-mmc fe088000.mmc: allocated mmc-pwrseq
> [    6.113160] mmc2: new UHS-I speed SDR104 SDIO card at address 0001
> [   19.206019] brcmfmac: brcmf_fw_alloc_request: using brcm/brcmfmac43456-sdio for chip BCM4345/9
> [   19.208638] brcmfmac mmc2:0001:1: Direct firmware load for brcm/brcmfmac43456-sdio.khadas,vim1s.bin failed with error -2
> [   19.218170] brcmfmac mmc2:0001:1: Direct firmware load for brcm/brcmfmac43456-sdio.clm_blob failed with error -2
> [   19.332091] brcmfmac: brcmf_c_process_clm_blob: no clm_blob available (err=-2), device may have limited channels available
> [   19.332113] brcmfmac: brcmf_c_process_txcap_blob: no txcap_blob available (err=-2)
> [   19.332327] brcmfmac: brcmf_c_preinit_dcmds: Firmware: BCM4345/9 wl0: Jun 25 2021 17:13:38 version 7.45.96.94 (7bb6b14@SYNA) (r745790) FWID 01-2bdfb64c es7.c5.n4.a3
> [   19.332731] brcmfmac: brcmf_sdio_read_control: read 232 control bytes failed: -84
> [   19.332988] brcmfmac: brcmf_sdio_rxfail: abort command, terminate frame, send NAK
> [   21.939090] brcmfmac: brcmf_sdio_bus_rxctl: resumed on timeout
> [   35.587422] meson-gx-mmc fe088000.mmc: unaligned sg len 96 blksize 512, disabling descriptor DMA for transfer

This is an issue with the driver, it's unrelated with the amlogic,dram-access-quirk, which fixes this because we use a bounce buffer.

> [   44.412677] ieee80211 phy0: brcmf_cfg80211_connect: BRCMF_C_SET_SSID failed (-52)
> [   45.696901] ieee80211 phy0: brcmf_cfg80211_connect: BRCMF_C_SET_SSID failed (-52)
> [   47.491768] ieee80211 phy0: brcmf_cfg80211_connect: BRCMF_C_SET_SSID failed (-52)
> [   53.292090] ieee80211 phy0: brcmf_cfg80211_connect: BRCMF_C_SET_SSID failed (-52)
> [   59.866328] ieee80211 phy0: brcmf_cfg80211_connect: BRCMF_C_SET_SSID failed (-52)
>
>
> Best regards,
> Nick